Chatbots seem to become more and more popular. The number of homepages offering chatbots is increasing, as is the number of posts explaining how to create a chatbot. This trend is not unfounded, and I think you can do some exciting things with chatbots.
With the growing trend towards chatbots, it is critical to understand how to create a successful and engaging chatbot. It’s also important to know what works and what doesn’t work when using chatbots. This article will give essential hints, tips and advice for anyone who wants to create a chatbot.
When you start your first chatbot project, don’t get knee-deep in the technical details. You should take a step back. Think about what to look for in such a project. What are the best practices when dealing with SAP Conversational AI and chatbots?
So what do you need to consider when creating a chatbot?
Have you ever asked yourself this question? Since we were confronted with this question, we did some research. In this article, we’d like to share with you what we’ve found out about best practices for bot creation to help you get your project started faster.
Fortunately, it doesn’t seem too difficult for both companies and budding chatbot developers to develop a high-quality chatbot, as long as you know what to look for in such a project.
Below, we would like to share with you a design pattern and the best and worst chatbot practices to help you get the most out of your investment and succeed with your chatbot.
Whether you’re planning a chatbot for your customer service or an engaging marketing bot for various messengers, knowing the principles of chatbot design and development will help ensure your chatbot effectively meets your customers’ expectations.
The most important things in a nutshell
Don’t you have much time? The most important aspects you should consider are quickly explained and obvious.
- Your customers should know that they are writing with a bot.
- Your chatbot should look charming and lively.
- The chatbot must have reliable answers to users’ questions.
- It is better if the bot is only very good at a few things than if it is not convincing at many things.
- Your customers should be able to forward the conversation to a person, and your chatbot should also proactively suggest this if it gets stuck.
Before we get to the individual best practices, we would like to briefly introduce you to a design pattern that you will encounter more often when searching for best practices in the SAP Conversational AI environment. The Receptionist Pattern. For example, our colleagues at SAP Conversational AI mention the pattern in this post about the best enterprise architectures for conversational interfaces.
The goal of any company is to provide excellent customer service that best meets customer expectations – nothing more, nothing less. Therefore, integrating a chatbot should not limit the range of topics. It should cover all areas.
The Receptionist Pattern is a bot design pattern that places the bot at the beginning of every user request via chat. Each question goes to the bot, which should then understand what the request refers to.
So the core is that the chatbot understands each request. However, this does not mean that the chatbot has to handle all requests itself. After understanding the question, the bot can either:
- Process the request independently in a fully automated conversation
- Start the conversation to collect important information, e.g. customer number, email address, and then hand over the conversation to a human agent
- Directly route the request to the correct human agent group depending on the subject
So the pattern is not vertical but horizontal. This architecture allows you to respond flexibly to any customer request. It doesn’t matter if you do this through a human or a bot, as long as the experience is smooth.
The bot works hand-in-hand with human agents for more in-depth or advanced questions by gathering customer information such as the contract number, the correct phone number, a specific bill number, or other information before passing the conversation to the agents.
This is where the strength of the interplay with different solutions becomes apparent. In combination with Sinch Contact Center, you could make handoffs to different agent groups with appropriately weighted agent capabilities.
Using the design pattern, SAP Conversational AI customers significantly increased customer service satisfaction by cutting call duration in half and reducing referral rates.
Similarly, customers saw a reduction in abandoned calls from 15% to 0%. Abandoned calls are those where a customer starts a call and then never contacts the agent again.
The background here is usually the long waiting time until an agent responds. Since there is no waiting time with a chatbot, the user is less likely to abandon the conversation. In addition, the chatbot does all the onboarding, so if the user does not engage with the conversation, it doesn’t route the chat to a human agent. As a result, employees waste less time on pointless requests and service quality across the organization and customer satisfaction increase.
For a successful, well-received chatbot, your customers need to know what your chatbot can do. This is especially true if your bot has severely limited functionality. Pay attention to the greeting when introducing your chatbot. The welcome text of your chatbot gives the user a first impression. This can influence further chat usage and the user’s overall satisfaction with the bot.
Your chatbot should introduce itself and give users an overview of its features. By giving customers explicit instructions on how to interact with your chatbot, you keep users on track. By listing its functions in advance, the chatbot prevents users from being disappointed and gives them ideas and instructions on interacting. By keeping customer expectations in mind from the start, you can provide a more satisfying experience for your users.
Make your users appreciate your chatbot and not have to wonder what purpose it serves. For example, provide specific question or answer choices. Users are comfortable. Sie akzeptieren sofortige Hilfe von Ihrem Chatbot, wenn der Chatbot das Problem lösen kann, anstatt auf einen Servicemitarbeiter zu warten. Gestalten Sie Ihren Bot benutzerfreundlich.
Show the customer answer options at appropriate points. Users prefer the straightforward way and click a button instead of typing their query themselves. You can also use buttons or examples to control better how your users use the bot. However, be careful about which platforms you choose to publish your bot – not all of them support buttons.
Regardless of the purpose of your chatbot and your design, your bot will always reach certain limits. The reason may be unplanned conversation flows or unavailable features. Regardless, it’s essential to explain to your customers what the bot can and cannot do. So, take the time to consider contingencies.
It’s best to explain to users what your chatbot can’t do and what it can. You can also redirect users to another question to keep them within the capabilities of your bot.
Offer only relevant content for your target audience. No one wants to deal with topics that don’t interest them.
There are many ways for your chatbot to deliver valuable and engaging content. Depending on the purpose and design of your chatbot, your chatbot could send images, links, GIFs, and emojis to break up information and make the conversation more interesting.
It’s more convenient for your users to receive multiple brief messages than one long message.
With chatbots, make sure your customers can easily talk to a human. If your chatbot takes over after business hours, let your customers know what times they can speak to an agent if they want to.
By enabling a handoff to a human, you ensure your chatbot doesn’t become a barrier between your customer and a satisfying user experience.
Depending on the environment, users will expect different things from your bot. If you place it in the public area of your homepage, a customer will probably accept that it needs to authenticate for specific actions. If users see a chatbot in an online service logged-in area, they will expect to know all the data about them already.
When creating your chatbot, consider and leverage the systems that are already in place in your organization. This will give your bot more functionality and make it more useful.
Be careful not to create extra work by missing integrations. Instead, pass the call directly to an agent at such points.
You should not assume that your chatbot will learn to handle all situations on its own. Therefore, you need to monitor your bot’s and users’ behavior and continuously adapt it to new questions.
Test the chatbot internally before making it available to customers. This will help reduce the number of unpredictable responses. Make sure that the testers don’t have too much knowledge about your chatbot in advance. It should be self-explanatory.
Once your chatbot goes live, review the messages you receive and adjust your bot regularly. Make sure you identify and fix gaps in the conversation flow or missing features that users expect on time.
Talk to a wide variety of people when training or designing your bot. Try to avoid having the same people interact with the bot all the time, as knowledge of the design and capabilities can lead to certain aspects being ignored. Try to have other people with different backgrounds interact with the bot and use their actions and reactions to improve the bot. Only a diverse group will understand the various facets of what your bot will encounter in an actual situation.
Platforms for creating chatbots are becoming more and more intuitive. It’s relatively easy to configure the bot and get it up and running quickly. However, before you jump the gun and neglect the bot’s design, consider the negative impact an immature chatbot can have on the user experience.
For example, a rushed chatbot in the service area can lead to lower customer satisfaction. More customers might call in for support, and your users will be wary of using your chatbot a second time. Take your time and carefully craft the overall design. You can develop the details over time as you learn more.
Often, external limitations are not as significant an obstacle to chatbot success as they seem. Although your chatbot provider should have a robust feature set and your interfaces should enable as many features as possible, it’s not just the technical capabilities of your bot that determine the quality of your chatbot. It’s also about the quality of your text and conversation flow.
The better the conversation, the more engaging and helpful your chatbot will be. This is true regardless of external constraints. The more conversational capabilities you add to your chatbot, the more powerful it will be. And the more information and similar questions you add, the better your chatbot can help your customers.
After your bot has introduced itself and communicated initial recommendations for action, it must respond to the user’s activities.
If your bot stops responding at any point, the user will feel that the chatbot is unreliable and will abort. Always try to provide a fallback option for the user and your bot to get out of such deadlocks.
Your users can change the topic of the conversation or ask a follow-up question. They might use sentence fragments, abbreviations, misspellings, and region-specific terminology. Make sure your chatbot is not too prone to resulting misunderstandings.
So don’t forget to consider multiple word choices, phrasing, and launch options when creating your bot, or you’ll end up with a frustratingly inflexible chatbot.
If possible, do not send large blocks of text to your users. These are difficult to read and can frustrate and overwhelm users.
Be sure to present information efficiently. Your chatbot can convey a lot of information – but the data should be broken down and offered in a way that is easy for the reader to digest. This motivates the user to continue interacting with the chatbot and increases satisfaction.
So keep the statements short and sweet. Don’t spam anyone. Do you have an extra long form? Consider if there is a better medium to answer. Some tasks are better done on a website.
How many times have you been on the phone and endured a useless automated response system (called an IVR) when all you wanted to do was talk to an employee?
A successful chatbot is a valuable tool, not a hindrance. Regardless of your chatbot’s function, don’t force your users (accidentally or intentionally) into a conversation with your bot. Allow your user to talk to a human and then route the chat to an agent.
A common but easily avoidable mistake in chatbot creation is to give sole responsibility for chatbot development to a single department, such as IT. This results in a functional chatbot in terms of its technology but lacks the expertise and knowledge that other departments could have brought to the table.
Developing a chatbot without HR can cause a bot whose personality doesn’t fit the company culture. Creating a chatbot without customer service can produce a bot that omits crucial frequently asked questions and does not handle user requests in a way that meets customer expectations. Developing a chatbot without a marketing or social media department can proceed in a chatbot that does not engage its audience through social channels.
When creating a chatbot, companies often try to build the functionality from scratch. This can cause an unreliable chatbot that doesn’t work well with the company’s existing systems.
For example, let’s say you’re creating a bot to collect meter readings from an energy provider’s customers. If your chatbot doesn’t communicate with existing systems, additional work for the company in handling requests that come in through this new channel. The captured meter readings would need to be received, checked for plausibility, and entered the system.
Bots are part of a larger ecosystem that includes multiple touchpoints between customers and brands. Creating a chatbot in a silo can be very damaging to both companies and customers.
Be clear from the start about which platforms you will publish your bot on. Buttons and images can be more accessible for some functions but are not available on some channels like Alexa Skills or SMS.
So think about text responses as a fallback option. This can also be faster and easier for users who already know the capabilities of the chatbot.
Don’t spend hours trying to create a stable flow of conversation by deliberately trying to outsmart your chatbot and find errors. At the latest, when the first customers chat with your first version of the bot, you’ll be surprised how many situations you didn’t expect. Close these gaps iteratively and make sure you have a suitable fallback mechanism from the beginning.
When developing chatbots, the goal is not always the same. Don’t start building your chatbot before you know what purpose it will serve.
If your goal for the chatbot is to save the company money and resources, consider using it to support a single department, such as customer service or HR. If your goal for the chatbot is to increase awareness of your brand, think about creating a useful, interesting chatbot that visitors can interact with.
The goal of your chatbot affects not only what your bot does but also where you should deploy it. If the goal is to provide customer support, consider deploying it on your website and mobile app. If your goal is marketing, then your chatbot will probably work best with a third-party messaging app, like Facebook Messenger.
If you don’t have a goal in mind for your chatbot, you risk developing a less effective bot on a less than ideal platform.
3 Take Aways
If you were to ask me what you should remember after reading this post, I would give you the following three points:
- Chatbots will take on a much more important role in a company’s customer service in the future, increasing service in a fun, relevant and efficient way!
- Perfecting your chatbot, like any new endeavor, requires time, patience, and a good attitude.
- By following these do’s and don’ts of chatbot design, you can avoid the most common chatbot mistakes and speed up your chatbot project.
Next steps to your own chatbot
Is your interest piqued? Look at some interesting and good examples from some SAP colleagues. Sebastien explains how to build your first chatbot in a Christoph explains how to build a chatbot with SAP Conversational AI in combination with S/4HANA Cloud.
Do you have experience with chatbots? What is your opinion and what do you think should be considered in such a project? Do you know of any other exciting resources to read?